This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH 1/3] nptl: Move nanosleep implementation to libc
* Adhemerval Zanella:
> On 06/11/2019 14:17, Florian Weimer wrote:
>> * Adhemerval Zanella:
>>
>>> On 06/11/2019 11:13, Florian Weimer wrote:
>>>> * Adhemerval Zanella:
>>>>
>>>>> I will check for all affected ABI, but I think there is no need. In fact
>>>>> I am not sure if Joseph rationale at b289cd9db8286fa6c670104dd5dfcfc68d5d00d6
>>>>> is currently true, we now need at least one absolute ABI symbol in at least
>>>>> one library.
>>>>
>>>> Hah, I thought I had broken the ABI test.
>>>>
>>>> I don't think the symbol matters, so that part is actually true, but we
>>>> do need to check the presence of the .gnu.version_d entry in a test
>>>> because that's what the loader expects (and as discussed, we do not want
>>>> to change that, at least not for lazy binding).
>>>>
>>>> Thanks,
>>>> Florian
>>>>
>>>
>>> Right, I checked the libpthread.so .gnu.version_d entries for every ABI
>>> affected and all of them contains the required versions (including for
>>> architectures which exports __nanosleep with a different version).
>>
>> Good to know, thanks.
>>
>> Florian
>>
>
> Ok with the __libpthread_version_placeholder change?
Yes, I think this is okay of the generated DSO looks reasonable.
> As a side note I will try to see if there is way to improve the
> .gnu.version_d entry with an automated test.
That would be nice!
Florian